<p>为此使用的任何解析器都需要是有状态的(即记住东西),而regex基本上不是有状态的。他们不适合做这项工作。在</p>
<p>如果这些是您唯一需要担心的数据类型,我想我应该为每个数据类型编写自定义解析器,在读取第一个字符之后将控制权传递给相应的解析器。在</p>
<p><strike>我现在真的要实施一个了,但是已经晚了。</strike></p>
<p><strong>好吧,我决定写一个实现:</strong></p>
<pre><code>from StringIO import StringIO
import string
inputs = ["10:a stringly",
"i1234e" ,
"l1:a1:be",
"d1:a1:b3:one3:twoe"]
# Constants
DICT_TYPE = 'd'
LIST_TYPE = 'l'
INT_TYPE = 'i'
TOKEN_EOF = ''
TOKEN_END = 'e'
COLON = ':'
class BadTypeIndicatorException(Exception):pass
def read_int(stream):
s = ""
while True:
ch = stream.read(1)
if ch not in [TOKEN_EOF, TOKEN_END, COLON]:
s += ch
else:
break
return s
def tokenize(stream):
s = ""
while True:
ch = stream.read(1)
if ch == TOKEN_END or ch == TOKEN_EOF:
return
if ch == COLON:
length = int(s)
yield stream.read(length)
s = ""
else:
s += ch
def parse(stream):
TYPE = stream.read(1)
if TYPE in string.digits:
length = int( TYPE + read_int(stream) )
return stream.read(length)
elif TYPE is INT_TYPE:
return int( read_int(stream) )
elif TYPE is LIST_TYPE:
return list(tokenize(stream))
elif TYPE is DICT_TYPE:
tokens = list(tokenize(stream))
return dict(zip(tokens[0::2], tokens[1::2]))
else:
raise BadTypeIndicatorException
for input in inputs:
stream = StringIO(input)
print parse(stream)
</code></pre>